We are integrating biotic information from the extensive Late Quaternary (with a particular focus on the last 25 thousand years) fossil record of temperate vertebrates and plants directly into the latest biodiversity-conservation models to improve predictions of species range shifts, better identify ecological traits that have made some species more (or less) prone to regional and range-wide extinctions, test the validity of threatened-species assessment approaches and pinpoint habitats that support the persistence of species and populations in response to shifting climates.
The spatial and temporal details of the fossil record, the occurrence of abrupt climate change events and the ability to connect past phenomena with ongoing, observable environmental processes and biological patterns means that the Late Quaternary period is a perfect time horizon for this type of research programme.
Fordham DA, Akcakaya HR, Alroy J, Saltré F, Wigley TM, Brook BW Predicting and mitigating future biodiversity loss using long-term ecological proxies. Nature Climate Change 6, 909–916 doi:10.1038/nclimate3086
Saltré F., Rodríguez-Rey M., Brook B.W., Johnson C.N., Turney C, Alroy J, Cooper A, Beeton N, Bird M.I., Fordham D.A. et al. Climate change not to blame for late Quaternary megafauna extinctions in Australia. Nature Communications 7, 10511 doi:10.1038/ncomms10511
Figure | Approaches for integrating long-term historical knowledge into ecological models. Methods that are able to combine the most recent developments in geochronology, palaeoclimate reconstructions, molecular biology and quantitative ecology (including palaeoecology) should improve: (1) predictions of biodiversity loss via independent validation in time and space, thereby strengthening our understanding of the ecological and environmental conditions required for more complex models; (2) knowledge of which ecological traits make some species more prone toextinction, thus allowing future conservation efforts to be better prioritized; (3) threatened species criteria used to categorize species according to extinction risk (for example, IUCN Red List or population viability analysis); and (4) predictions of the location of refugia and understanding of their ecological functioning, advancing the future conservation value of protected area networks. See Fordham et al. 2016 doi:10.1038/nclimate3086
Figure | Identifying the traits most likely to influence extinction risk and range dynamics using a robust coverage of a multi-dimensional parameter space. Spatial demographic models are built using ‘best estimates’ for demographic and environmental attributes, using information from congeneric species and allometry. Model parameters are varied across wide but plausible ranges using a robust coverage of multi-dimensional parameter space (Latin hypercube sampling), producing thousands of conceivable models with varying rates of population growth, dispersal, niche lability, human exploitation and so on. Model simulations of change in spatial abundance patterns are validated against independent fossil estimates of timing of extinction and genetic estimates of the expansion and contraction of Ne using approximate Bayesian computation (ABC) and machine learning techniques, for example, boosted regression trees (BRTs). These statistical methods can be used to determine model parameters that best predict timing of extinction or relative change in abundance. See Fordham et al. 2016 doi:10.1038/nclimate3086